# A tibble: 2 × 4
model .metric .estimator .estimate
<chr> <chr> <chr> <dbl>
1 Full roc_auc binary 0.875
2 SegSumT roc_auc binary 0.806
A model that randomly assigns classes to the data is expected to achieve an AUC of 0.5 (dotted line on the previous plot) while a perfect model would achieve an AUC of 1.
# A tibble: 2 × 4
model .metric .estimator .estimate
<chr> <chr> <chr> <dbl>
1 Full pr_auc binary 0.583
2 SegSumT pr_auc binary 0.447
A model that randomly assigns classes to the data is expected to achieve an PR-AUC of # successes / n while a perfect model would achieve an PR-AUC of 1 (a point at a coordinate of (1,1)).